Remy Klaassen

Feasibility and repeatability of PET with the hypoxia tracer [18F]HX4 in oesophageal and pancreatic cancer

BACKGROUND AND PURPOSE To investigate the feasibility and to determine the repeatability of recurrent [(18)F]HX4 PET scans in patients with oesophageal (EC) and pancreatic (PC) cancer. MATERIALS AND METHODS 32 patients were scanned in total; seven patients (4 EC/3 PC) were scanned 2, 3 and 4h post injection (PI) of [(18)F]HX4 and 25 patients (15 EC/10 PC) were scanned twice 3.5h PI, on two separate days (median 4, range 1-9days). Maximum tumour to background ratio (TBRmax) and the tumour hypoxic volume (HV) (TBR>1.0) were calculated. Repeatability was assessed using Bland-Altman analysis. Agreement in localization was calculated as the distance between the centres of mass in the HVs. RESULTS For EC, the TBRmax in the tumour (mean±SD) was 1.87±0.46 with a coefficient of repeatability (CoR) of 0.53 (28% of mean). The HV ranged from 3.4 to 98.8ml with a CoR of 5.1ml. For PC, the TBRmax was 1.72±0.23 with a CoR of 0.27 (16% of mean). The HV ranged from 4.6 to 104.0ml with a CoR of 7.8ml. The distance between the centres of mass in the HV was 2.2±1.3mm for EC and 2.1±1.5mm for PC. CONCLUSIONS PET scanning with [(18)F]HX4 was feasible in both EC and PC patients. Amount and location of elevated [(18)F]HX4 uptake showed good repeatability, suggesting [(18)F]HX4 PET could be a promising tool for radiation therapy planning and treatment response monitoring in EC and PC patients.

Minimizing the Acquisition Time for Intravoxel Incoherent Motion Magnetic Resonance Imaging Acquisitions in the Liver and Pancreas

ObjectivesThe aim of this study was to determine the combination of b-values and signal averages for diffusion-weighted image acquisitions that render the minimum acquisition time necessary to obtain values of the intravoxel incoherent motion (IVIM) model parameters in vivo in the pancreas or liver with acceptable reproducibility. Materials and MethodsFor 16 volunteers, diffusion-weighted images, with 14 b-values and 9 acquisitions per b-value, were acquired in 2 scan sessions. The IVIM model was fitted to data from lesion-sized regions of interest (ROIs) (1.7 cm3) as well as organ-sized ROIs in the pancreas and liver. By deleting data during analyzes, the IVIM model parameters, D and f, could be determined as a function of the number of b-values as well as the number of measurements per b-value taken along. For the IVIM model parameters, we examined the behavior reproducibility, in the form of the within-subject coefficient of variation (CVw), as a function of the amount of data taken along in the fits. Finally, we determined the minimum acquisition time required as a function of CVw. ResultFor the lesion-sized ROI, the intersession CVws were 8%/46% and 13%/55% for D/f in the pancreas and liver, respectively, when all data were taken along. For 1.2 times larger CVws, acquisition in the pancreas could be done in 5:15 minutes using 9 acquisitions per b-value at b = 0, 30, 50, 65, 100, 375, and 500 mm−2s and for the liver in 2:15 using 9 acquisitions per b-value at b = 0, 40, and 500 mm−2s. ConclusionsAcquiring 7 b-values in the pancreas and 3 b-values in the liver only decreases the reproducibility by 20% compared with an acquisition with 14 b-values. The understanding of the behavior of reproducibility as a function of b-values and acquisitions per b-values scanned will help researchers select the shortest IVIM protocol.

Addition of MRI for CT-based pancreatic tumor delineation: a feasibility study

Abstract Purpose: To assess the effect of additional magnetic resonance imaging (MRI) alongside the planning computed tomography (CT) scan on target volume delineation in pancreatic cancer patients. Material and methods: Eight observers (radiation oncologists) from six institutions delineated the gross tumor volume (GTV) on 3DCT, and internal GTV (iGTV) on 4DCT of four pancreatic cancer patients, while MRI was available in a second window (CT + MRI). Variations in volume, generalized conformity index (CIgen), and overall observer variation, expressed as standard deviation (SD) of the distances between delineated surfaces, were analyzed. CIgen is a measure of overlap of the delineated iGTVs (1 = full overlap, 0 = no overlap). Results were compared with those from an earlier study that assessed the interobserver variation by the same observers on the same patients on CT without MRI (CT-only). Results: The maximum ratios between delineated volumes within a patient were 6.1 and 22.4 for the GTV (3DCT) and iGTV (4DCT), respectively. The average (root-mean-square) overall observer variations were SD = 0.41 cm (GTV) and SD = 0.73 cm (iGTV). The mean CIgen was 0.36 for GTV and 0.37 for iGTV. When compared to the iGTV delineated on CT-only, the mean volumes of the iGTV on CT + MRI were significantly smaller (32%, Wilcoxon signed-rank, p < .0005). The median volumes of the iGTV on CT + MRI were included for 97% and 92% in the median volumes of the iGTV on CT. Furthermore, CT + MRI showed smaller overall observer variations (root-mean-square SD = 0.59 cm) in six out of eight delineated structures compared to CT-only (root-mean-square SD = 0.72 cm). However, large local observer variations remained close to biliary stents and pathological lymph nodes, indicating issues with instructions and instruction compliance. Conclusions: The availability of MRI images during target delineation of pancreatic cancer on 3DCT and 4DCT resulted in smaller target volumes and reduced the interobserver variation in six out of eight delineated structures.

Quantitative assessment of biliary stent artifacts on MR images: Potential implications for target delineation in radiotherapy

PURPOSE Biliary stents may cause susceptibility artifacts, gradient-induced artifacts, and radio frequency (RF) induced artifacts on magnetic resonance images, which can hinder accurate target volume delineation in radiotherapy. In this study, the authors investigated and quantified the magnitude of these artifacts for stents of different materials. METHODS Eight biliary stents made of nitinol, platinum-cored nitinol, stainless steel, or polyethylene from seven vendors, with different lengths (57-98 mm) and diameters (3.0-11.7 mm), were placed in a phantom. To quantify the susceptibility artifacts sequence-independently, ΔB0-maps and T2∗-maps were acquired at 1.5 and 3 T. To study the effect of the gradient-induced artifacts at 3 T, signal decay in images obtained with maximum readout gradient-induced artifacts was compared to signal decay in reference scans. To quantify the RF induced artifacts at 3 T, B1-maps were acquired. Finally, ΔB0-maps and T2∗-maps were acquired at 3 T of two pancreatic cancer patients who had received platinum-cored nitinol biliary stents. RESULTS Outside the stent, susceptibility artifacts dominated the other artifacts. The stainless steel stent produced the largest susceptibility artifacts. The other stents caused decreased T2∗ up to 5.1 mm (1.5 T) and 8.5 mm (3 T) from the edge of the stent. For sequences with a higher bandwidth per voxel (1.5 T: BWvox > 275 Hz/voxel; 3 T: BWvox > 500 Hz/voxel), the B0-related susceptibility artifacts were negligible (<0.2 voxels). The polyethylene stent showed no artifacts. In vivo, the changes in B0 and T2∗ induced by the stent were larger than typical variations in B0 and T2∗ induced by anatomy when the stent was at an angle of 30° with the main magnetic field. CONCLUSIONS Susceptibility artifacts were dominating over the other artifacts. The magnitudes of the susceptibility artifacts were determined sequence-independently. This method allows to include additional safety margins that ensure target irradiation.

Revisiting the Potential of Alternating Repetition Time Balanced Steady-State Free Precession Imaging of the Abdomen at 3 T.

ObjectiveThe aim was to investigate the value of optimized 3-dimensional alternating repetition time balanced steady-state free precession (ATR-SSFP), as an alternative to conventional segmented balanced steady-state free precession (bSSFP) with fat suppression prepulse (FS-bSSFP), in single breath-hold abdominal magnetic resonance imaging at 3 T. MethodsBloch simulations were performed to determine the optimal flip angle (FA = 1–90 degrees) and &tgr; (1–3) with respect to signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) between abdominal organs for ATR-SSFP. These were corroborated by phantom measurements for different T1/T2 values (5–47) as well as in a healthy volunteer. In addition, fat suppression efficiency was studied using phantom and volunteer measurements. The effect of resolution on image quality was studied in a healthy volunteer. Using the optimal settings, ATR-SSFP images as well as FS-bSSFP images were obtained in 15 pancreatic cancer patients. For 10 structures of interest, the signal ratio with respect to the pancreas was computed and compared between both sequences. Finally, 10 items on image quality (fat suppression, artifacts, and sharpness) and tissue conspicuity (ducts, vessels, and duodenum) were scored by 2 abdominal radiologists for both image sequences. ResultsThe results of simulations, phantom measurements, and volunteer measurements showed that, considering scan time, fat suppression, and clinical relevance, the ideal settings for ATR-SSFP were as follows: &tgr; = 3; TR1 = 3.46 milliseconds; radiofrequency phase cycling 0, 180, 180, 0 degrees; and FA = 13–16 degrees (highest SNR) and 24–26 degrees (highest CNR). The optimized feasible additional settings implemented for patient scans were FA = 18 degrees and resolution = 1.4 × 1.4 × 1.4 mm3. In patients, the signal ratios of both ATR-SSFP and FS-bSSFP were comparable and had a T2-like contrast behavior, although more accentuated in ATR-SSFP. The ATR-SSFP scored significantly higher than FS-bSSFP for 9 of 10 items scored. ConclusionsFor single breath-hold abdominal imaging at 3 T, ATR-SSFP performs best with &tgr; = 3 and an FA between 13 degrees (highest SNR) and 26 degrees (highest CNR). The scoring of both abdominal radiologists indicated that, at &tgr; = 3, FA = 18 degrees, and 1.4 × 1.4 × 1.4 mm3 resolution, ATR-SSFP was preferred over conventional FS-bSSFP with similar settings.

Comparison of six fit algorithms for the intra-voxel incoherent motion model of diffusion-weighted magnetic resonance imaging data of pancreatic cancer patients

The intravoxel incoherent motion (IVIM) model for diffusion-weighted imaging (DWI) MRI data bears much promise as a tool for visualizing tumours and monitoring treatment response. To improve the currently poor precision of IVIM, several fit algorithms have been suggested. In this work, we compared the performance of two Bayesian IVIM fit algorithms and four other IVIM fit algorithms for pancreatic cancer imaging. DWI data were acquired in 14 pancreatic cancer patients during two MRI examinations. Three different measures of performance of the fitting algorithms were assessed: (i) uniqueness of fit parameters (Spearman’s rho); (ii) precision (within-subject coefficient of variation, wCV); and (iii) contrast between tumour and normal-appearing pancreatic tissue. For the diffusivity D and perfusion fraction f, a Bayesian fit (IVIM-Bayesian-lin) offered the best trade-off between tumour contrast and precision. With the exception for IVIM-Bayesian-lin, all algorithms resulted in a very poor precision of the pseudo-diffusion coefficient D* with a wCV of more than 50%. The pseudo-diffusion coefficient D* of the Bayesian approaches were, however, significantly correlated with D and f. Therefore, the added value of fitting D* was considered limited in pancreatic cancer patients. The easier implemented least squares fit with fixed D* (IVIM-fixed) performed similar to IVIM-Bayesian-lin for f and D. In conclusion, the best performing IVIM fit algorithm was IVM-Bayesian-lin, but an easier to implement least squares fit with fixed D* performs similarly in pancreatic cancer patients.

Repeatability and correlations of dynamic contrast enhanced and T2* MRI in patients with advanced pancreatic ductal adenocarcinoma

BACKGROUND In current oncological practice of pancreatic ductal adenocarcinoma (PDAC), there is a great demand for response predictors and markers for early treatment evaluation. In this study, we investigated the repeatability and the interaction of dynamic contrast enhanced (DCE) and T2* MRI in patients with advanced PDAC to enable for such evaluation using these techniques. MATERIALS & METHODS 15 PDAC patients underwent two DCE, T2* and anatomical 3 T MRI sessions before start of treatment. Parametric maps were calculated for the transfer constant (Ktrans), rate constant (kep), extracellular extravascular space (ve) and perfusion fraction (vp). Quantitative R2* (1/T2*) maps were obtained from the multi-echo T2* images. Differences between normal and cancerous pancreas were determined using a Wilcoxon matched pairs test. Repeatability was obtained using Bland-Altman analysis and relations between DCE and T2*/R2* were observed by Spearman correlation and voxel-wise binned plots of tumor voxels. RESULTS PDAC Ktrans (p = 0.007), kep (p < 0.001), vp (p = 0.035) were lower and ve (p < 0.001) was higher compared to normal pancreas. The coefficient of variation between sessions was 21.8% for Ktrans, 9.9% for kep, 19.3% for ve, 18.2% for vp and 18.7% for R2*. Variation between patients ranged from 20.2% for kep to 43.6% for Ktrans. In the tumor both Ktrans (r = 0.56, p = 0.030) and ve (r = 0.54, p = 0.037) showed a positive correlation with T2*. Voxel wise analysis showed a steep increase in R2* for tumor voxels with lower Ktrans and ve. CONCLUSION We showed good repeatability of DCE and T2* related MRI parameters in advanced PDAC patients. Furthermore, we have illustrated the relation of DCE Ktrans and ve with tissue T2* and R2* indicating substantial value of these parameters for detecting tumor hypoxia in future studies. The results from our study pave the way for further response evaluation studies and patient selection based on DCE and T2* parameters.

TU-H-206-08: Quantitative Impact of Biliary Stent Artefacts On MR Images

PURPOSE Biliary stents may cause artifacts in MR-images. These artifacts are caused by differences in susceptibility (artifact-1), eddy currents during read-out (artifact-2) and eddy currents due to the excitation pulse (artifact-3), which all are MRI sequence-dependent. In this study, we investigated and quantified the magnitude of these artifacts as a function of the distance from biliary stents, using a sequence-independent approach where possible. METHODS Eight phantoms were made containing different commercial biliary stents made of nitinol, platinol, stainless steel or polyethylene. To quantify artifact-1 sequence-independently, we acquired ΔB0-maps and T2*-maps. To maximize artifact-2, the phantom was placed 130mm off center. We then compared signal ratios from acquisitions with gradients in directions that induced eddy currents and acquisitions with gradients in directions that did not induce eddy currents, to assess the contribution of artifact-2. We acquired B1-maps to quantify the effect of artifact-3.These experiments were performed at 3T. The ΔB0-maps and T2*-maps were also acquiered at 1.5T.Finally, we acquired ΔB0-maps, T2*-maps and clinical MR-images in vivo in two pancreatic cancer patients with platinol stents. RESULTS Artifact-1 was dominant over the other artifacts. The stainless steel stent showed the largest artifact-1: decreased T2* and ΔB0>75Hz up to 20mm from the stents edge which results in signal shifts>3voxels for typical diffusion weighted images with bandwidth BW=25Hz/voxel. The other stents showed a decreased T2* up to 8.5mm/5.1mm (3T/1.5T) from the edge of the stent and had a BW<95Hz/55Hz, resulting in signal shifts <0.2voxels for sequences with BW>500Hz/275Hz at 3T/1.5T. The plastic stent showed no artifacts. In vivo, the changes in B0 and T2* induced by the stent were larger than typical variations in B0 and T2* induced by anatomy. CONCLUSION We presented a sequence-independent measure to quantify susceptibility artifacts. Other artifacts are negligible compared to susceptibility artifacts for biliary stents.

Abstract 421: The role of the tumor microenvironment of pancreatic cancer to predict treatment outcome

Introduction: Pancreatic ductal adenocarcinoma (PDAC) ranks 4th on the list of cancer-related deaths which is mostly due to therapy resistance. The tumor microenvironment is one of the main hallmarks to contribute to therapy resistance in which tumors are characterized by large amounts of stroma deposition, and hypovascularization. In this study we assess the use of non-invasive functional magnetic resonance imaging (MRI) for characterization of these tumor microenvironment characteristics and its prognostic and predictive role. Diffusion weighted (DWI) MRI and dynamic contrast enhanced (DCE)-MRI provide insight in stromal content and tumor vascularization. Subsequent correlation of imaging with immunohistochemical (IHC) stainings can validate the biological basis of these imaging modalities. Here, we propose a method to correlate functional MRI to corresponding histology derived tissue characteristics. Methods: Patients suffering from (borderline) resectable PDAC underwent DWI and DCE-MRI on a 3T Philips scanner prior to resection and/or neoadjuvant chemoradiation. DWI data were fitted with the intra voxel incoherent motion model to obtain, among others, the diffusion coefficient D, while DCE-MRI data were fitted according to the extended Tofts model to retrieve the volume transfer constant Ktrans. After resection, the tissue was sectioned in the axial plane and slices where photographed. One slice with evident tumor was selected for whole mount embedding. Whole mount sections were stained for stromal content (αSMA) and vascular density (CD31) The pancreas contour of the photographed tissue slices was manually outlined. Slices were aligned and stacked to form a 3D volume of the pancreas specimen, which was projected onto the MR images. Stained whole mount sections were aligned with the corresponding tissue photograph and transposed to the MRI accordingly to correlate pathology sections to imaging sections and validate parameters using IHC. Results: Sixteen patients were included in the study of which four patients have currently been analyzed. Tissue specimens after resection showed good visual agreement with MRI and could be matched using corresponding anatomical landmarks. In addition, Lower D-values, suggesting higher cell density, indeed corresponded with higher stromal content, while higher values on the Ktrans map, suggesting a higher tumor vascularity, corresponded with higher vessel density. Conclusion: With these preliminary data we demonstrated the feasibility of matching post-operative tissue slices to in vivo MRI of the pancreas. Matching stained whole mount sections to MRI images provides direct information on localization and local tumor heterogeneity for validation of functional MRI parameters. Expanding this work in a larger patient group will provide valuable insights on the underlying biology of functional MRI parameters in pancreatic cancer. Citation Format: Anne Steins, Remy Klaassen, Oliver Gurney-Champion, Maarten Bijlsma, Jan Paul Medema, Hessel Wijkstra, Geertjan van Tienhoven, Olivier Busch, Cornelis Punt, Marc Besselink, Hanneke Wilmink, Marc van de Vijver, Jaap Stoker, Aart Nederveen, Hanneke van Laarhoven. The role of the tumor microenvironment of pancreatic cancer to predict treatment outcome. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 421. doi:10.1158/1538-7445.AM2015-421